Professional Documents
Culture Documents
Corporate Headquarters
Cisco Systems, Inc.
170 West Tasman Drive
San Jose, CA 95134-1706
USA
http://www.cisco.com
Tel: 408 526-4000
800 553-NETS (6387)
Fax: 408 526-4100
THE SOFTWARE LICENSE AND LIMITED WARRANTY FOR THE ACCOMPANYING PRODUCT ARE SET FORTH IN THE INFORMATION PACKET THAT
SHIPPED WITH THE PRODUCT AND ARE INCORPORATED HEREIN BY THIS REFERENCE. IF YOU ARE UNABLE TO LOCATE THE SOFTWARE LICENSE
OR LIMITED WARRANTY, CONTACT YOUR CISCO REPRESENTATIVE FOR A COPY.
The following information is for FCC compliance of Class A devices: This equipment has been tested and found to comply with the limits for a Class A digital device, pursuant
to part 15 of the FCC rules. These limits are designed to provide reasonable protection against harmful interference when the equipment is operated in a commercial
environment. This equipment generates, uses, and can radiate radio-frequency energy and, if not installed and used in accordance with the instruction manual, may cause
harmful interference to radio communications. Operation of this equipment in a residential area is likely to cause harmful interference, in which case users will be required
to correct the interference at their own expense.
The following information is for FCC compliance of Class B devices: The equipment described in this manual generates and may radiate radio-frequency energy. If it is not
installed in accordance with Cisco’s installation instructions, it may cause interference with radio and television reception. This equipment has been tested and found to
comply with the limits for a Class B digital device in accordance with the specifications in part 15 of the FCC rules. These specifications are designed to provide reasonable
protection against such interference in a residential installation. However, there is no guarantee that interference will not occur in a particular installation.
Modifying the equipment without Cisco’s written authorization may result in the equipment no longer complying with FCC requirements for Class A or Class B digital
devices. In that event, your right to use the equipment may be limited by FCC regulations, and you may be required to correct any interference to radio or television
communications at your own expense.
You can determine whether your equipment is causing interference by turning it off. If the interference stops, it was probably caused by the Cisco equipment or one of its
peripheral devices. If the equipment causes interference to radio or television reception, try to correct the interference by using one or more of the following measures:
• Move the equipment to one side or the other of the television or radio.
• Plug the equipment into an outlet that is on a different circuit from the television or radio. (That is, make certain the equipment and the television or radio are on circuits
controlled by different circuit breakers or fuses.)
Modifications to this product not authorized by Cisco Systems, Inc. could void the FCC approval and negate your authority to operate the product.
The Cisco implementation of TCP header compression is an adaptation of a program developed by the University of California, Berkeley (UCB) as part of UCB’s public
domain version of the UNIX operating system. All rights reserved. Copyright © 1981, Regents of the University of California.
NOTWITHSTANDING ANY OTHER WARRANTY HEREIN, ALL DOCUMENT FILES AND SOFTWARE OF THESE SUPPLIERS ARE PROVIDED “AS IS” WITH
ALL FAULTS. CISCO AND THE ABOVE-NAMED SUPPLIERS DISCLAIM ALL WARRANTIES, EXPRESSED OR IMPLIED, INCLUDING, WITHOUT
LIMITATION, THOSE OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OR ARISING FROM A COURSE OF
DEALING, USAGE, OR TRADE PRACTICE.
IN NO EVENT SHALL CISCO OR ITS SUPPLIERS BE LIABLE FOR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, OR INCIDENTAL DAMAGES, INCLUDING,
WITHOUT LIMITATION, LOST PROFITS OR LOSS OR DAMAGE TO DATA ARISING OUT OF THE USE OR INABILITY TO USE THIS MANUAL, EVEN IF CISCO
OR ITS SUPPLIERS HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
CCSP, CCVP, the Cisco Square Bridge logo, Follow Me Browsing, and StackWise are trademarks of Cisco Systems, Inc.; Changing the Way We Work, Live, Play, and Learn, and
iQuick Study are service marks of Cisco Systems, Inc.; and Access Registrar, Aironet, BPX, Catalyst, CCDA, CCDP, CCIE, CCIP, CCNA, CCNP, Cisco, the Cisco Certified
Internetwork Expert logo, Cisco IOS, Cisco Press, Cisco Systems, Cisco Systems Capital, the Cisco Systems logo, Cisco Unity, Enterprise/Solver, EtherChannel, EtherFast,
EtherSwitch, Fast Step, FormShare, GigaDrive, GigaStack, HomeLink, Internet Quotient, IOS, IP/TV, iQ Expertise, the iQ logo, iQ Net Readiness Scorecard, LightStream,
Linksys, MeetingPlace, MGX, the Networkers logo, Networking Academy, Network Registrar, Packet, PIX, Post-Routing, Pre-Routing, ProConnect, RateMUX, ScriptShare,
SlideCast, SMARTnet, The Fastest Way to Increase Your Internet Quotient, and TransPath are registered trademarks of Cisco Systems, Inc. and/or its affiliates in the United States
and certain other countries.
All other trademarks mentioned in this document or Website are the property of their respective owners. The use of the word partner does not imply a partnership relationship
between Cisco and any other company. (0601R)
Any Internet Protocol (IP) addresses used in this document are not intended to be actual addresses. Any examples, command display output, and figures included in the
document are shown for illustrative purposes only. Any use of actual IP addresses in illustrative content is unintentional and coincidental.
Preface vii
Attenuation 2-8
Dispersion 2-10
Summary 2-12
SONET/SDH 3-1
ATM 3-1
IP 3-2
FDDI 3-4
INDEX
This Introduction to DWDM Technology is intended for readers who want to gain a basic understanding
of dense wavelength division multiplexing (DWDM) technology and its role in metropolitan area
networks (MANs). The contents are organized as follows:
Additional Reading
Readers who are interested in more detailed information about optical networking will find that there is
a wide selection of resources available, especially on SONET. The following publications are good,
basic texts:
• Dutton, Harry J. R. Understanding Optical Communications. Research Triangle Park: IBM
Corporation; 1998.
• Goff, David R. Fiber Optic Reference Guide, 2nd edition. Boston: Focal Press; 1999.
• Goralski, Walter J. SONET, 2nd edition. New York: McGraw-Hill; 2000.
• Kaminow, Ivan P., and Koch, Thomas L., eds. Optical Fiber Telecommunications. San Diego:
Academic Press; 1997.
The following publication is a detailed text on DWDM, particularly from the standpoint of theory and
optical components:
Kartalopoulos, Stamatios V. Introduction to DWDM Technology: Data in a Rainbow. New York:
IEEE Press; 1999.
Although comprehensive texts specifically about DWDM are not yet plentiful, there is a variety of
information in the form of tutorials, white papers, and so on, to be found on the World Wide Web.
For definitions of terms and acronyms used in this Introduction, refer to the Glossary of Optical
Networking Terms.
The following discussion provides some background on why dense wavelength division multiplexing
(DWDM) is an important innovation in optical networks and what benefits it can provide. We begin with
a high-level view of the segments of the global network and the economic forces driving the revolution
in fiber optic networks. We then examine the differences between traditional time-division multiplexing
(TDM) and wavelength division multiplexing (WDM). Finally, we explore the advantages of this new
technology.
Long-haul
networks
Mesh Ring
Point-to-point
Metropolitan
Intra-metro Intra-metro networks
connections Inter-metro connections
connections
Enterprise:
Residential networks: SONET: ISP POP: - Frame Relay Access
- DSL - POTS -IP services - ESCON networks
- Cable modem - Wireless - Fibre Channel
49367
- Gigabit Ethernet
Long-Haul Networks
Long-haul networks are at the core of the global network. Dominated by a small group of large
transnational and global carriers, long-haul networks connect the MANs. Their application is transport,
so their primary concern is capacity. In many cases these networks, which have traditionally been based
on Synchronous Optical Network (SONET) or Synchronous Digital Hierarchy (SDH) technology, are
experiencing fiber exhaust as a result of high bandwidth demand.
Access Networks
At the other end of the spectrum are the access networks. These networks are the closest to the end users,
at the edge of the MAN. They are characterized by diverse protocols and infrastructures, and they span
a broad spectrum of rates. Customers range from residential Internet users to large corporations and
institutions. The predominance of IP traffic, with its inherently bursty, asymmetric, and unpredictable
nature, presents many challenges, especially with new real-time applications. At the same time, these
networks are required to continue to support legacy traffic and protocols, such as IBM’s Enterprise
System Connection (ESCON).
An Alternative View
The preceding breakdown of the global network represents a somewhat simplified view. In reality, the
lines between the domains are not always so clear-cut. Long-haul and metropolitan networks are
sometimes not clearly delineated; the same holds true for the access and metropolitan domains.
Furthermore, other views of the global network exist. One, for example, defines the access network as
part of, rather than separate from, the MAN, while also including enterprise connectivity in the
metropolitan domain. In this view, the metropolitan market breaks down as follows:
• Core—These are essentially scaled-down long-haul systems. They are considered the core of the
MAN, because they interconnect carrier POPs and do not directly interface with end users.
• Metropolitan access—This is the segment between carrier POPs and access facilities, which could
be equipment at customer premises or at an aggregation point.
• Enterprise—This is the part of the network dedicated to serving the needs of enterprises. Using
owned or leased fiber (or leased fiber capacity), connectivity is provided between geographically
disparate enterprise sites and for new applications, such as storage area networks (SANs).
Economic Forces
As we enter the twenty-first century, it goes without saying that information services have permeated
society and commerce. Information, while still a tool, has become a commodity in itself. Yet the
universal acceptance and ubiquitous adoption of information technology systems has strained the
backbones on which they were built. High demand—coupled with high usage rates, a deregulated
telecommunications environment, and high availability requirements—is rapidly depleting the
capacities of fibers that, when installed 10 years ago, were expected to suffice for the foreseeable future.
Bandwidth Demand
The explosion in demand for network bandwidth is largely due to the growth in data traffic, specifically
Internet Protocol (IP). Leading service providers report bandwidths doubling on their backbones about
every six to nine months. This is largely in response to the 300 percent growth per year in Internet traffic,
while traditional voice traffic grows at a compound annual rate of only about 13 percent (see Figure 1-2).
Data
250
Traffic volume
200
150
Voice
100
50
Voice-centric Data-centric
48066
At the same time that network traffic volume is increasing, the nature of the traffic itself is becoming
more complex. Traffic carried on a backbone can originate as circuit based (TDM voice and fax), packet
based (IP), or cell based (ATM and Frame Relay). In addition, there is an increasing proportion of delay
sensitive data, such as voice over IP and streaming video.
In response to this explosive growth in bandwidth demand, along with the emergence of IP as the
common foundation for all services, long-haul service providers are moving away from TDM based
systems, which were optimized for voice but now prove to be costly and inefficient. Meanwhile,
metropolitan networks are also experiencing the impact of growing congestion, as well as rapidly
changing requirements that call for simpler and faster provisioning than is possible with older equipment
and technologies. Of key importance in the metropolitan area is the growth in storage area networks
(SANs), discussed in the “Storage Area Networks” section on page 3-5.
Bandwidth = n
No protection
Bandwidth = n/2
48067
Dedicated protection
Time-Division Multiplexing
Time-division multiplexing (TDM) was invented as a way of maximizing the amount of voice traffic that
could be carried over a medium. In the telephone network before multiplexing was invented, each
telephone call required its own physical link. This proved to be an expensive and unscalable solution.
Using multiplexing, more than one telephone call could be put on a single link.
TDM can be explained by an analogy to highway traffic. To transport all the traffic from four tributaries
to another city, you can send all the traffic on one lane, providing the feeding tributaries are fairly
serviced and the traffic is synchronized. So, if each of the four feeds puts a car onto the trunk highway
every four seconds, then the trunk highway would get a car at the rate of one each second. As long as
the speed of all the cars is synchronized, there would be no collision. At the destination the cars can be
taken off the highway and fed to the local tributaries by the same synchronous mechanism, in reverse.
This is the principle used in synchronous TDM when sending bits over a link. TDM increases the
capacity of the transmission link by slicing time into smaller intervals so that the bits from multiple input
sources can be carried on the link, effectively increasing the number of bits transmitted per second (see
Figure 1-4).
Transmission link
...
48069
With TDM, input sources are serviced in round-robin fashion. Though fair, this method results in
inefficiency, because each time slot is reserved even when there is no data to send. This problem is
mitigated by the statistical multiplexing used in Asynchronous Transfer Mode (ATM). Although ATM
offers better bandwidth utilization, there are practical limits to the speed that can be achieved due to the
electronics required for segmentation and reassembly (SAR) of ATM cells that carry packet data.
(T)
48070
The original unit used in multiplexing telephone calls is 64 kbps, which represents one phone call.
Twenty-four (in North America) or thirty-two (outside North America) of these units are multiplexed
using TDM into a higher bit-rate signal with an aggregate speed of 1.544 Mbps or 2.048 Mbps for
transmission over T1 or E1 lines, respectively.The hierarchy for multiplexing telephone calls is shown
in Table 1-1.
These are the basic building blocks used by SONET/SDH to multiplex into a standard hierarchy of
speeds, from STS-1 at 51.85 Mbps to STS-192/STM-64 at 10 Gbps. Table 1-2 shows the relationship
between the telco signal rates and the most commonly used levels of the SONET/SDH hierarchy
(OC-768 is not yet common).
Figure 1-6 depicts this multiplexing and aggregation hierarchy. Using a standard called virtual
tributaries for mapping lower-speed channels into the STS-1 payload, the 28 DS1 signals can be mapped
into the STS-1 payload, or they can be multiplexed to DS3 with an M13 multiplexer and fit directly into
the STS-1. Note also that ATM and Layer 3 traffic, using packet over SONET (POS), can feed into the
SONET terminal from switches equipped with SONET interfaces.
DS1
1
VT STS-1
mapping
1
28
Multiplexed
signal DS1 STS-1
DS3
M13 SONET terminal
STSx
24 or 28 Unchannelized multiplexing To optical
30 STS-3C components
ATM switch
ATM cells
Router/Layer 3
switch
IP datagrams (Packet over SONET)
48071
SONET/SDH does have some drawbacks. As with any TDM, the notions of priority or congestion do not
exist in SONET or SDH. Also, the multiplexing hierarchy is a rigid one. When more capacity is needed,
a leap to the next multiple must be made, likely resulting in an outlay for more capacity than is initially
needed. For example, the next incremental step from 10 Gbps (STS-192) TDM is 40 Gbps (STS-768).
Also, since the hierarchy is optimized for voice traffic, there are inherent inefficiencies when carrying
data traffic with SONET frames. Some of these inefficiencies are shown in Table 1-3. DWDM, by
contrast, can transport any protocol, including SONET, without special encapsulation.
To summarize, the demand placed on the transport infrastructure by bandwidth-hungry applications and
the explosive growth of the Internet has exceeded the limits of traditional TDM. Fiber, which once
promised seemingly unlimited bandwidth, is being exhausted, and the expense, complexity, and
scalability limitations of the SONET infrastructure are becoming increasingly problematic.
WDM
48072
Note The term wavelength is used instead of the term frequency to avoid confusion with other
uses of frequency. Wavelength is often used interchangeably with lambda and channel.
In a WDM system, each of the wavelengths is launched into the fiber, and the signals are demultiplexed
at the receiving end. Like TDM, the resulting capacity is an aggregate of the input signals, but WDM
carries each input signal independently of the others. This means that each channel has its own dedicated
bandwidth; all signals arrive at the same time, rather than being broken up and carried in time slots.
The difference between WDM and dense wavelength division multiplexing (DWDM) is fundamentally
one of only degree. DWDM spaces the wavelengths more closely than does WDM, and therefore has a
greater overall capacity.The limits of this spacing are not precisely known, and have probably not been
reached, though systems are available in mid-year 2000 with a capacity of 128 lambdas on one fiber.
DWDM has a number of other notable features, which are discussed in greater detail in the following
chapters. These include the ability to amplify all the wavelengths at once without first converting them
to electrical signals, and the ability to carry signals of different speeds and types simultaneously and
transparently over the fiber (protocol and bit rate independence).
Note WDM and DWDM use single-mode fiber to carry multiple lightwaves of differing
frequencies. This should not be confused with transmission over multimode fiber, in which
light is launched into the fiber at different angles, resulting in different “modes” of light. A
single wavelength is used in multimode transmission.
Mix of
DS1 Channel 1
DS3
OC-1 Single fiber
OC-3 TDM
(one wavelength)
OC-12
OC-48 Channel n
Mix of
OC-12
OC-48
OC-192 Single fiber
SONET (multiple WDM
ATM wavelengths)
GE
48073
Additional Drivers in Metropolitan Area Networks
Bandwidth, the chief driver in the long-haul market, is also a big driver in metropolitan area, access, and
large enterprise networks (see Figure 1-9). In these types of networks additional applications driving
demand for bandwidth include storage area networks (SANs), which make possible the serverless office,
consolidation of data centers, and real-time transaction processing backup.
Today Future
T1 OC-3 OC-12
T3 OC-48
Fractional T1
ISP or ISP or
enterprise enterprise
WAN WAN/MAN
T1/T3 OC-12
OC-48
OC-192
GE
48068
There is also rapidly increasing demand on access networks, which function primarily to connect end
users over low-speed connections, such as dial-up lines, DSL, cable, and wireless, to a local POP. These
connections are typically aggregated and carried over a SONET ring, which at some point attaches to a
local POP that serves as an Internet gateway for long hauls. Now, the growing demand for high-speed
services is prompting service providers to transform the POP into a dynamic service-delivery center. As
a result, it is increasingly likely that a customer now obtains many high-speed services directly from the
POP, without ever using the core segment of the Internet.
The need to provision services of varying types in a rapid and efficient manner in response to the
changing demands of customers is a distinguishing characteristic of the metropolitan networks. With
SONET, which is the foundation of the vast majority of existing MANs, service provisioning is a lengthy
and complex process. Network planning and analysis, ADM provisioning, Digital Crossconnect System
(DCS) reconfiguration, path and circuit verification, and service creation can take several weeks. By
contrast, with DWDM equipment in place provisioning new service can be as simple as turning on
another lightwave in an existing fiber pair.
Potential providers of DWDM-based services in metropolitan areas, where abundant fiber plant already
exists or is being built, include incumbent local exchange carriers (ILECs), competitive local exchange
carriers (CLECs), inter-exchange carriers (IXCs), Internet service providers (ISPs), cable companies,
private network operators, and utility companies. Such carriers can often offer new services for less cost
than older ones. Much of the cost savings is due to reducing unnecessary layers of equipment, which also
lowers operational costs and simplifies the network architecture.
Carriers can create revenue today by providing protocol-transparent, high-speed LAN and SAN services
to large organizations, as well as a mixture of lower-speed services (Token Ring, FDDI, Ethernet) to
smaller organizations. In implementing an optical network, they are ensuring that they can play in the
competitive field of the future.
Why DWDM?
From both technical and economic perspectives, the ability to provide potentially unlimited transmission
capacity is the most obvious advantage of DWDM technology. The current investment in fiber plant can
not only be preserved, but optimized by a factor of at least 32. As demands change, more capacity can
be added, either by simple equipment upgrades or by increasing the number of lambdas on the fiber,
without expensive upgrades. Capacity can be obtained for the cost of the equipment, and existing fiber
plant investment is retained.
Bandwidth aside, DWDM’s most compelling technical advantages can be summarized as follows:
• Transparency—Because DWDM is a physical layer architecture, it can transparently support both
TDM and data formats such as ATM, Gigabit Ethernet, ESCON, and Fibre Channel with open
interfaces over a common physical layer.
• Scalability—DWDM can leverage the abundance of dark fiber in many metropolitan area and
enterprise networks to quickly meet demand for capacity on point-to-point links and on spans of
existing SONET/SDH rings.
• Dynamic provisioning—Fast, simple, and dynamic provisioning of network connections give
providers the ability to provide high-bandwidth services in days rather than months.
In the following sections we discuss some additional advantages, including migration from SONET and
reliability.
SONET WDM
terminal
OC-n Fiber
OC-48c
ATM
switch OC-48c
Router/
Layer 3
48074
switch
Optical signals become attenuated as they travel through fiber and must be periodically regenerated in
core networks. In SONET/SDH optical networks prior to the introduction of DWDM, each separate fiber
carrying a single optical signal, typically at 2.5 Gbps, required a separate electrical regenerator every 60
to 100 km (37 to 62 mi). As additional fibers were “turned up” in a core network, the total cost of
regenerators could become very large, because not only the cost of the regenerators themselves, but also
the facilities to house and power them, had to be considered. The need to add regenerators also increased
the time required to light new fibers
The upper part of Figure 1-11 shows the infrastructure required to transmit at 10 Gbps (4 x OC-48 SR
interfaces) across a span of 360 km (223 mi) using SONET equipment; the lower part of the figure shows
the infrastructure required for the same capacity using DWDM. While optical amplifiers could be used
in the SONET case to extend the distance of spans before having to boost signal power, there would still
need to be an amplifier for each fiber. Because with DWDM all four signals can be transported on a
single fiber pair (versus four), fewer pieces of equipment are required. Eliminating the expense of
regenerators (RPTR) required for each fiber results in considerable savings.
40km 40km 40km 40km 40km 40km 40km 40km 40km 40km
1310 1310 1310 1310 1310 1310 1310 1310 1310
TERM TERM
1310
RPTR RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310
TERM TERM
RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310
TERM TERM
RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310 RPTR1310
TERM TERM
RPTR RPTR RPTR RPTR RPTR RPTR RPTR RPTR RPTR
48075
A single optical amplifier can reamplify all the channels on a DWDM fiber without demultiplexing and
processing them individually, with a cost approaching that of a single regenerator. The optical amplifier
merely amplifies the signals; it does not reshape, retime or retransmit them as a regenerator does, so the
signals may still need to be regenerated periodically. But depending on system design, signals can now
be transmitted anywhere from 600 to thousands of kilometers without regeneration.
In addition to dramatically reducing the cost of regenerators, DWDM systems greatly simplify the
expansion of network capacity. The only requirement is to install additional or higher bit-rate interfaces
in the DWDM systems at either end of the fiber. In some cases it will only be necessary to increase the
number of lambdas on the fiber by deploying existing interfaces, as shown in the upper half of
Figure 1-12. The existing optical amplifiers amplify the new channel without additional regenerators. In
the case of adding higher bit-rate interfaces, as shown in the lower half of Figure 1-12, fiber type can
become a consideration. See the “Optical Fibers” section on page 2-5 for an overview of types of optical
fibers and their uses.
OC-48 OC-48
Add more wavelengths
OC-48 OC-48
DWDM
DWDM
OC-48 OC-48
DWDM
2.5 Gb/s 2.5 Gbps
10 Gb/s 10 Gbps
2.5 Gb/s 2.5 Gbps
48076
10 Gb/s 10 Gbps
Although amplifiers are of great benefit in long-haul transport, they are often unnecessary in
metropolitan networks. Where distances between network elements are relatively short, signal strength
and integrity can be adequate without amplification. But with MANs expanding in deeper into long-haul
reaches, amplifiers will become useful.
Additional Benefits
The shift in the makeup of traffic from voice to data has important implications for the design and
operation of carrier networks. The introduction of cell-switching technologies such as ATM and Frame
Relay demonstrates the limitations of the narrow-band, circuit-switched network design, but the limits
of these technologies are being reached. Data is no longer an add-on to the voice-centric network, but is
central. There are fundamentally different requirements of a data-centric network; two of these are the
aggregation model and the open versus proprietary interfaces.
Aggregation in a voice-centric network consists of multiplexing numerous times onto transmission
facilities and at many points in the network. Aggregation in a data-centric network, by contrast, tends to
happen at the edge. With OC-48 (and higher) interfaces readily available on cell and packet switches, it
becomes possible to eliminate costly SONET multiplexing and digital cross-connect equipment. OC-48
connections can interface directly to DWDM equipment.
Finally, service providers and enterprises can respond more quickly to changing demands by allocating
bandwidth on demand. The ability to provision services rapidly by providing wavelength on demand
creates new revenue opportunities such as wavelength leasing (an alternative to leasing of physical links
or bit rate-limited tunnels), disaster recovery, and optical VPNs.
The emergence of DWDM is one of the most recent and important phenomena in the development of
fiber optic transmission technology. In the following discussion we briefly trace the stages of fiber optic
technology and the place of DWDM in that development. We then examine the functions and
components of a DWDM system, including the enabling technologies, and conclude with a high-level
description of the operation of a DWDM system.
of its lower attenuation, followed by a third window (C band) at 1550 nm with an even lower optical loss.
Today, a fourth window (L band) near 1625 nm is under development and early deployment. These four
windows are shown relative to the electromagnetic spectrum in Figure 2-1.
700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 nm
48077
850 1310 1550 1625
Send
End End
system system
(1310 nm + 850 nm)
Receive
End End
48078
system system
(1310 nm + 850 nm)
The early 1990s saw a second generation of WDM, sometimes called narrowband WDM, in which two
to eight channels were used. These channels were now spaced at an interval of about 400 GHz in the
1550-nm window. By the mid-1990s, dense WDM (DWDM) systems were emerging with 16 to 40
channels and spacing from 100 to 200 GHz. By the late 1990s DWDM systems had evolved to the point
where they were capable of 64 to 160 parallel channels, densely packed at 50 or even 25 GHz intervals.
As Figure 2-3 shows, the progression of the technology can be seen as an increase in the number of
wavelengths accompanied by a decrease in the spacing of the wavelengths. Along with increased density
of wavelengths, systems also advanced in their flexibility of configuration, through add-drop functions,
and management capabilities.
Evolution of
DWDM
64+ channels
Late 25-50 GHz spacing
1990's
16+ channels 100-200 GHz spacing
1996 Dense WDM, integrated systems
with network management, add-drop functions.
2-8 channels
Early
1990's Passive WDM
200-400 GHz spacing
Passive WDM components/parts
2 channels
1980's Wideband WDM
48079
1310 nm, 1550 nm
Increases in channel density resulting from DWDM technology have had a dramatic impact on the
carrying capacity of fiber. In 1995, when the first 10 Gbps systems were demonstrated, the rate of
increase in capacity went from a linear multiple of four every four years to four every year (see
Figure 2-4).
1000000
Mbps
1000000
100000
10000
1000
100
10
48080
1979 1981 1983 1985 1987 1989 1990 1991 1993 1995 1997 1998
Transmitters Receivers
Combining Separating
signals signals
Transmission on fiber
48081
The system performs the following main functions:
• Generating the signal—The source, a solid-state laser, must provide stable light within a specific,
narrow bandwidth that carries the digital data, modulated as an analog signal.
• Combining the signals—Modern DWDM systems employ multiplexers to combine the signals.
There is some inherent loss associated with multiplexing and demultiplexing. This loss is dependent
upon the number of channels but can be mitigated with optical amplifiers, which boost all the
wavelengths at once without electrical conversion.
• Transmitting the signals—The effects of crosstalk and optical signal degradation or loss must be
reckoned with in fiber optic transmission. These effects can be minimized by controlling variables
such as channel spacings, wavelength tolerance, and laser power levels. Over a transmission link,
the signal may need to be optically amplified.
• Separating the received signals—At the receiving end, the multiplexed signals must be separated
out. Although this task would appear to be simply the opposite of combining the signals, it is
actually more technically difficult.
• Receiving the signals—The demultiplexed signal is received by a photodetector.
In addition to these functions, a DWDM system must also be equipped with client-side interfaces to
receive the input signal. This function is performed by transponders (see the “Interfaces to DWDM”
section on page 2-20). On the DWDM side are interfaces to the optical fiber that links DWDM systems.
Enabling Technologies
Optical networking, unlike SONET/SDH, does not rely on electrical data processing. As such, its
development is more closely tied to optics than to electronics. In its early form, as described previously,
WDM was capable of carrying signals over two widely spaced wavelengths, and for a relatively short
distance. To move beyond this initial state, WDM needed both improvements in existing technologies
and invention of new technologies. Improvements in optical filters and narrowband lasers enabled
DWDM to combine more than two signal wavelengths on a fiber. The invention of the flat-gain optical
amplifier, coupled in line with the transmitting fiber to boost the optical signal, dramatically increased
the viability of DWDM systems by greatly extending the transmission distance.
Other technologies that have been important in the development of DWDM include improved optical
fiber with lower loss and better optical transmission characteristics, EDFAs, and devices such as fiber
Bragg gratings used in optical add/drop multiplexers.
Optical Fibers
The following discussion of DWDM components and technologies includes a refresher on optical fibers,
with emphasis on their application for DWDM. Background information on subjects such as the
properties of light and optical theory can be found in many readily available printed sources and online,
for example, in the tutorial at http://www.vislab.usyd.edu.au/photonics/fibres/index.html.
• The incident angle is less than the critical angle. The critical angle is the angle of incidence at which
light stops being refracted and is instead totally reflected.
The principle of total internal reflection within a fiber core is illustrated in Figure 2-6. The core has a
higher refractive index than the cladding, allowing the beam that strikes that surface at less than the
critical angle to be reflected. The second beam does not meet the critical angle requirement and is
refracted.
n2 Cladding
n1
Reflected Core
48082
n = index of refraction
n1 > n2 gives total internal reflection
An optical fiber consists of two different types of highly pure, solid glass (silica)—the core and the
cladding—that are mixed with specific elements, called dopants, to adjust their refractive indices. The
difference between the refractive indices of the two materials causes most of the transmitted light to
bounce off the cladding and stay within the core. The critical angle requirement is met by controlling the
angle at which the light is injected into the fiber. Two or more layers of protective coating around the
cladding ensure that the glass can be handled without damage.
Cladding
Core
48083
To compensate for the dispersion drawback of step-index multimode fiber, graded-index fiber was
invented. Graded-index refers to the fact that the refractive index of the core is graded—it gradually
decreases from the center of the core outward. The higher refraction at the center of the core slows the
speed of some light rays, allowing all the rays to reach their destination at about the same time and
reducing modal dispersion.
The second general type of fiber, single-mode, has a much smaller core that allows only one mode of
light at a time through the core (see Figure 2-8). As a result, the fidelity of the signal is better retained
over longer distances, and modal dispersion is greatly reduced. These factors attribute to a higher
bandwidth capacity than multimode fibers are capable of. For its large information-carrying capacity and
low intrinsic loss, single-mode fibers are preferred for longer distance and higher bandwidth
applications, including DWDM.
Cladding
Core
48084
single-mode (SM) fibers, were designed for use in the second window, near 1310 nm. To optimize the
fiber’s performance in this window, the fiber was designed so that chromatic dispersion would be close
to zero near the 1310-nm wavelength.
As optical fiber use became more common and the needs for greater bandwidth and distance increased,
a third window, near 1550 nm, was exploited for single-mode transmission. The third window, or C band,
offered two advantages: it had much lower attenuation, and its operating frequency was the same as that
of the new erbium-doped fiber amplifiers (EDFAs). However, its dispersion characteristics were severely
limiting. This was overcome to a certain extent by using narrower linewidth and higher power lasers. But
because the third window had lower attenuation than the 1310-nm window, manufacturers came up with
the dispersion-shifted fiber design, which moved the zero-dispersion point to the 1550-nm region.
Although this solution now meant that the lowest optical attenuation and the zero-dispersion points
coincided in the 1550-nm window, it turned out that there are destructive nonlinearities in optical fiber
near the zero-dispersion point for which there is no effective compensation. Because of this limitation,
these fibers are not suitable for DWDM applications.
The third type, non-zero dispersion-shifted fiber, is designed specifically to meet the needs of DWDM
applications. The aim of this design is to make the dispersion low in the 1550-nm region, but not zero.
This strategy effectively introduces a controlled amount of dispersion, which counters nonlinear effects
such as four-wave mixing (see the “Other Nonlinear Effects” section on page 2-11) that can hinder the
performance of DWDM systems.
Transmission Challenges
Transmission of light in optical fiber presents several challenges that must be dealt with. These fall into
the following three broad categories:
• Attenuation—decay of signal strength, or loss of light power, as the signal propagates through the
fiber
• Chromatic dispersion—spreading of light pulses as they travel down the fiber
• Nonlinearities—cumulative effects from the interaction of light with the material through which it
travels, resulting in changes in the lightwave and interactions between lightwaves
Each of these effects has several causes, not all of which affect DWDM. The discussion in the following
sections addresses those causes that are relevant to DWDM.
Attenuation
Attenuation in optical fiber is caused by intrinsic factors, primarily scattering and absorption, and by
extrinsic factors, including stress from the manufacturing process, the environment, and physical
bending. The most common form of scattering, Rayleigh scattering, is caused by small variations in the
density of glass as it cools. These variations are smaller than the wavelengths used and therefore act as
scattering objects (see Figure 2-9). Scattering affects short wavelengths more than long wavelengths and
limits the use of wavelengths below 800 nm.
48085
Attenuation due to absorption is caused by the intrinsic properties of the material itself, the impurities
in the glass, and any atomic defects in the glass. These impurities absorb the optical energy, causing the
light to become dimmer (see Figure 2-10). While Rayleigh scattering is important at shorter
wavelengths, intrinsic absorption is an issue at longer wavelengths and increases dramatically above
1700 nm. However, absorption due to water peaks introduced in the fiber manufacturing process are
being eliminated in some new fiber types.
48086
The primary factors affecting attenuation in optical fibers are the length of the fiber and the wavelength
of the light. Figure 2-11 shows the loss in decibels per kilometer (dB/km) by wavelength from Rayleigh
scattering, intrinsic absorption, and total attenuation from all causes.
5
First window
3
Optical loss
(dB/km)
Total
2 attenuation
Intrinsic
1
Intrinsic scattering absorbtion
(Rayleigh
48087
0
0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
Wavelength (micrometers)
Attenuation in fiber is compensated primarily through the use of optical amplifiers, as described in the
“Optical Amplifiers” section on page 2-15.
Dispersion
Dispersion is the spreading of light pulses as they travel down optical fiber. Dispersion results in
distortion of the signal (see Figure 2-12), which limits the bandwidth of the fiber.
48088
Time Fiber Time
Two general types of dispersion affect DWDM systems. One of these effects, chromatic dispersion, is
linear while the other, polarization mode dispersion (PMD), is nonlinear.
Chromatic Dispersion
Chromatic dispersion occurs because different wavelengths propagate at different speeds. The effect of
chromatic dispersion increases as the square of the bit rate. In single-mode fiber, chromatic dispersion
has two components, material dispersion and waveguide dispersion.
Material dispersion occurs when wavelengths travel at different speeds through the material. A light
source, no matter how narrow, emits several wavelengths within a range. Thus, when this range of
wavelengths travels through a medium, each individual wavelength arrives at a different time.
The second component of chromatic dispersion, waveguide dispersion, occurs because of the different
refractive indices of the core and the cladding of fiber. The effective refractive index varies with
wavelength as follows:
• At short wavelengths, the light is well confined within the core. Thus the effective refractive index
is close to the refractive index of the core material.
• At medium wavelengths, the light spreads slightly into the cladding. This decreases the effective
refractive index.
• At long wavelengths, much of the light spreads into the cladding. This brings the effective refractive
index very close to that of the cladding.
This result of the phenomenon of waveguide dispersion is a propagation delay in one or more of the
wavelengths relative to others.
Total chromatic dispersion, along with its components, is plotted by wavelength in Figure 2-13 for
dispersion-shifted fiber. For non-dispersion-shifted fiber, the zero dispersion wavelength is 1310 nm.
30 Material
dispersion
20 component
-10
Wavelength
-20 dispersion
component
-30
-40
-50
48089
Wavelength (nm)
Though chromatic dispersion is generally not an issue at speeds below OC-48, it does increase with
higher bit rates due to the spectral width required. New types of zero-dispersion-shifted fibers greatly
reduce these effects. The phenomenon can also be mitigated with dispersion compensators.
Most single-mode fibers support two perpendicular polarization modes, a vertical one and a horizontal
one. Because these polarization states are not maintained, there occurs an interaction between the pulses
that results is a smearing of the signal.
Polarization mode dispersion (PMD) is caused by ovality of the fiber shape as a result of the
manufacturing process or from external stressors. Because stress can vary over time, PMD, unlike
chromatic dispersion, is subject to change over time. PMD is generally not a problem at speeds below
OC-192.
In addition to PMD, there are other nonlinear effects. Because nonlinear effects tend to manifest
themselves when optical power is very high, they become important in DWDM.
Linear effects such as attenuation and dispersion can be compensated, but nonlinear effects accumulate.
They are the fundamental limiting mechanisms to the amount of data that can be transmitted in optical
fiber. The most important types of nonlinear effects are stimulated Brillouin scattering, stimulated
Raman scattering, self-phase modulation, and four-wave mixing. In DWDM, four-wave mixing is the
most critical of these types.
Four-wave mixing is caused by the nonlinear nature of the refractive index of the optical fiber. Nonlinear
interactions among different DWDM channels creates sidebands that can cause interchannel
interference. In Figure 2-14 three frequencies interact to produce a fourth frequency, resulting in
cross-talk and signal-to-noise degradation.
Power
f1 ffwm f3 f2
48090
Frequencies
The effect of four-wave mixing is to limit the channel capacity of a DWDM system. Four-wave mixing
cannot be filtered out, either optically or electrically, and increases with the length of the fiber. Due to
its propensity for four-wave-mixing, DSF is unsuitable for WDM applications. This prompted the
invention of NZ-DSF, which takes advantage of the fact that a small amount of chromatic dispersion can
be used to mitigate four-wave mixing.
Summary
In the long-distance network, the majority of embedded fiber is standard single-mode (G.652) with high
dispersion in the 1550-nm window, which limits the distance for OC-192 transmission. Dispersion can
be mitigated to some extent, and at some cost, using dispersion compensators. Non-zero
dispersion-shifted fiber can be deployed for OC-192 transport, but higher optical power introduces
nonlinear effects.
In the short-haul network, PMD and nonlinear effects are not so critical as they are in long-haul systems,
where higher speeds (OC-192 and higher) are more common. DWDM systems using optical signals of
2.5 Gbps or less are not subject to these nonlinear effects at short distances.
The major types of single-mode fibers and their application can be summarized as follows:
• Non-dispersion-shifted fiber (standard SM fiber)—accounts for greater than 95 percent of deployed
plant; suitable for TDM (single-channel) use in the 1310-nm region or DWDM use in the 1550-nm
region (with dispersion compensators). This type of fiber can also support 10 Gigabit Ethernet
standard at distances over 300 meters.
• Dispersion-shifted fiber—suitable for TDM use in the 1550-nm region, but unsuitable for DWDM
in this region.
• Non-zero dispersion-shifted fiber—good for both TDM and DWDM use in the 1550-nm region.
• Newer generation fibers—includes types that allow the energy to travel further into the cladding,
creating a small amount of dispersion to counter four-wave mixing, and dispersion-flattened fibers,
which permit use of wavelengths farther from the optimum wavelength without pulse spreading.
Note As bit rates increase to 40 Gbps and beyond, the interdependence between system design
and fiber design will become increasingly important for strategic planning.
Note Monochromatic is a relative term; in practice there are only light sources within a certain
range. Stability of a light source is a measure of how constant its intensity and wavelength
are.
Two general types of light emitting devices are used in optical transmission, light-emitting diodes
(LEDs) and laser diodes, or semiconductor lasers. LEDs are relatively slow devices, suitable for use at
speeds of less than 1 Gbps, they exhibit a relatively wide spectrum width, and they transmit light in a
relatively wide cone. These inexpensive devices are often used in multimode fiber communications.
Semiconductor lasers, on the other hand, have performance characteristics better suited to single-mode
fiber applications.
Figure 2-15 shows the general principles of launching laser light into fiber. The laser diode chip emits
light in one direction to be focused by the lens onto the fiber and in the other direction onto a photodiode.
The photodiode, which is angled to reduce back reflections into the laser cavity, provides a way of
monitoring the output of the lasers and providing feedback so that adjustments can be made.
Lens
Fiber
Laser diode chip
Rear-facet
48091
monitor photodiode
Requirements for lasers include precise wavelength, narrow spectrum width, sufficient power, and
control of chirp (the change in frequency of a signal over time). Semiconductor lasers satisfy nicely the
first three requirements. Chirp, however, can be affected by the means used to modulate the signal.
In directly modulated lasers, the modulation of the light to represent the digital data is done internally.
With external modulation, the modulation is done by an external device. When semiconductor lasers are
directly modulated, chirp can become a limiting factor at high bit rates (above 10 Gbps). External
modulation, on the other hand, helps to limit chirp. The external modulation scheme is depicted in
Figure 2-16.
Electrical signal
DC current drive
External
modulator
ITU λ Modulated
48092
15xx distributed feedback laser ITU λ
Two types of semiconductor lasers are widely used, monolithic Fabry-Perot lasers, and distributed
feedback (DFB) lasers. The latter type is particularly well suited for DWDM applications, as it emits a
nearly monochromatic light, is capable of high speeds, has a favorable signal-to-noise ratio, and has
superior linearity. DFB lasers also have center frequencies in the region around 1310 nm, and from
1520 to 1565 nm. The latter wavelength range is compatible with EDFAs. There are many other types
and subtypes of lasers. Narrow spectrum tunable lasers are available, but their tuning range is limited to
approximately 100-200 GHz. Under development are wider spectrum tunable lasers, which will be
important in dynamically switched optical networks.
ITU Grid
Cooled DFB lasers are available in precisely selected wavelengths. The ITU draft standard G.692 defines
a laser grid for point-to-point WDM systems based on 100-GHz wavelength spacings with a center
wavelength of 1553.52 nm (see Table 2-1).
While this grid defines a standard, users are free to use the wavelengths in arbitrary ways and to choose
from any part of the spectrum. In addition, manufacturers can deviate from the grid by extending the
upper and lower bounds or by spacing the wavelengths more closely, typically at 50 GHz, to double the
number of channels. The closer the spacing, the more channel crosstalk results. In addition, the impact
of some fiber nonlinearities, such as FWM, increases. Spacing at 50 GHz also limits the maximum data
rate per wavelength to 10 Gbps. The implications of the flexibility in implementation are twofold: There
is no guarantee of compatibility between two end systems from different vendors, and there exists a
design trade-off in the spacing of wavelengths between number of channels and maximum bit rate.
Light Detectors
On the receive end, it is necessary to recover the signals transmitted at different wavelengths on the fiber.
Because photodetectors are by nature wideband devices, the optical signals are demultiplexed before
reaching the detector.
Two types of photodetectors are widely deployed, the positive-intrinsic-negative (PIN) photodiode and
the avalanche photodiode (APD). PIN photodiodes work on principles similar to, but in the reverse of,
LEDs. That is, light is absorbed rather than emitted, and photons are converted to electrons in a 1:1
relationship. APDs are similar devices to PIN photodiodes, but provide gain through an amplification
process: One photon acting on the device releases many electrons. PIN photodiodes have many
advantages, including low cost and reliability, but APDs have higher receive sensitivity and accuracy.
However, APDs are more expensive than PIN photodiodes, they can have very high current
requirements, and they are temperature sensitive.
Optical Amplifiers
Due to attenuation, there are limits to how long a fiber segment can propagate a signal with integrity
before it has to be regenerated. Before the arrival of optical amplifiers (OAs), there had to be a repeater
for every signal transmitted, as discussed earlier and shown in Figure 1-11. The OA has made it possible
to amplify all the wavelengths at once and without optical-electrical-optical (OEO) conversion. Besides
being used on optical links, optical amplifiers also can be used to boost signal power after multiplexing
or before demultiplexing, both of which can introduce loss into the system.
Erbium-doped
fiber (10-50 m)
Pump Pump
48093
laser laser
The key performance parameters of optical amplifiers are gain, gain flatness, noise level, and output
power. EDFAs are typically capable of gains of 30 dB or more and output power of +17 dB or more. The
target parameters when selecting an EDFA, however, are low noise and flat gain. Gain should be flat
because all signals must be amplified uniformly. While the signal gain provided with EDFA technology
is inherently wavelength-dependent, it can be corrected with gain flattening filters. Such filters are often
built into modern EDFAs.
Low noise is a requirement because noise, along with signal, is amplified. Because this effect is
cumulative, and cannot be filtered out, the signal-to-noise ratio is an ultimate limiting factor in the
number of amplifiers that can be concatenated and, therefore, the length of a single fiber link. In practice,
signals can travel for up to 120 km (74 mi) between amplifiers. At longer distances of 600 to 1000 km
(372 to 620 mi) the signal must be regenerated. That is because the optical amplifier merely amplifies
the signals and does not perform the 3R functions (reshape, retime, retransmit). EDFAs are available for
the C-band and the L-band.
In a unidirectional system (see Figure 2-18), there is a multiplexer at the sending end and a demultiplexer
at the receiving end. Two system would be required at each end for bidirectional communication, and
two separate fibers would be needed.
MUX DEMUX
DEMUX MUX
51183
In a bidirectional system, there is a multiplexer/demultiplexer at each end (see Figure 2-19) and
communication is over a single fiber pair.
MUX/DEMUX MUX/DEMUX
51184
Multiplexers and demultiplexers can be either passive or active in design. Passive designs are based on
prisms, diffraction gratings, or filters, while active designs combine passive devices with tunable filters.
The primary challenges in these devices is to minimize cross-talk and maximize channel separation.
Cross-talk is a measure of how well the channels are separated, while channel separation refers to the
ability to distinguish each wavelength.
n1
n2 > n1
λ1, λ2 λn
λ1
Lens B C λ2
Prism
Fiber Lens λ3
λn
Fibers at the
48095
focal points
Another technology is based on the principles of diffraction and of optical interference. When a
polychromatic light source impinges on a diffraction grating (see Figure 2-21), each wavelength is
diffracted at a different angle and therefore to a different point in space. Using a lens, these wavelengths
can be focused onto individual fibers.
Fibers Diffracted
λ1 wavelengths
λ2
λn
Lens
49172
Defraction
grating
Arrayed waveguide gratings (AWGs) are also based on diffraction principles. An AWG device,
sometimes called an optical waveguide router or waveguide grating router, consists of an array of
curved-channel waveguides with a fixed difference in the path length between adjacent channels (see
Figure 2-22). The waveguides are connected to cavities at the input and output. When the light enters the
input cavity, it is diffracted and enters the waveguide array. There the optical length difference of each
waveguide introduces phase delays in the output cavity, where an array of fibers is coupled. The process
results in different wavelengths having maximal interference at different locations, which correspond to
the output ports.
W1 Array of waveguides
WN
Array of fibers
λ1 + λ2 + ... λN
λ1
51036
λN
A different technology uses interference filters in devices called thin film filters or multilayer
interference filters. By positioning filters, consisting of thin films, in the optical path, wavelengths can
be sorted out (demultiplexed). The property of each filter is such that it transmits one wavelength while
reflecting others. By cascading these devices, many wavelengths can be demultiplexed (see
Figure 2-23).
Multiwavelength light
Multilayer
interference
filter
Demultiplexed
wavelengths
48096
Of these designs, the AWG and thin film interference filters are gaining prominence. Filters offer good
stability and isolation between channels at moderate cost, but with a high insertion loss. AWGs are
polarization-dependent (which can be compensated), and they exhibit a flat spectral response and low
insertion loss. A potential drawback is that they are temperature sensitive such that they may not be
practical in all environments. Their big advantage is that they can be designed to perform multiplexing
and demultiplexing operations simultaneously. AWGs are also better for large channel counts, where the
use of cascaded thin film filters is impractical.
λ1 = λ2 = λn
Fiber Fiber
λ1 = λ2 = λn λ1 = λ2 = λn
Amp OADM Amp
48097
λn λ n
There are two general types of OADMs. The first generation is a fixed device that is physically
configured to drop specific predetermined wavelengths while adding others. The second generation is
reconfigurable and capable of dynamically selecting which wavelengths are added and dropped.
Thin-film filters have emerged as the technology of choice for OADMs in current metropolitan DWDM
systems because of their low cost and stability. For the emerging second generation of OADMs, other
technologies, such as tunable fiber gratings and circulators, will come into prominence.
Interfaces to DWDM
Most DWDM systems support standard SONET/SDH short-reach optical interfaces to which any
SONET/SDH compliant client device can attach. In today's long-haul WDM systems, this is most often
an OC-48c/STM-16c interface operating at the 1310-nm wavelength. In addition, other interfaces
important in metropolitan area and access networks are commonly supported: Ethernet (including Fast
Ethernet and Gigabit Ethernet), ESCON, Sysplex Timer and Sysplex Coupling Facility Links, and Fibre
Channel. The new 10 Gigabit Ethernet standard is supported using a very short reach (VSR) OC-192
interface over MM fiber between 10 Gigabit Ethernet and DWDM equipment.
On the client side there can be SONET/SDH terminals or ADMs, ATM switches, or routers. By
converting incoming optical signals into the precise ITU-standard wavelengths to be multiplexed,
transponders are currently a key determinant of the openness of DWDM systems.
Within the DWDM system a transponder converts the client optical signal from back to an electrical
signal and performs the 3R functions (see Figure 2-25). This electrical signal is then used to drive the
WDM laser. Each transponder within the system converts its client's signal to a slightly different
wavelength. The wavelengths from all of the transponders in the system are then optically multiplexed.
In the receive direction of the DWDM system, the reverse process takes place. Individual wavelengths
are filtered from the multiplexed fiber and fed to individual transponders, which convert the signal to
electrical and drive a standard interface to the client.
48098
equipment
Future designs include passive interfaces, which accept the ITU-compliant light directly from an
attached switch or router with an optical interface.
Terminal A Terminal B
Transponder Transponder
interfaces interfaces
7. The incoming signal is demultiplexed into individual DWDM lambdas (or wavelengths).
8. The individual DWDM lambdas are mapped to the required output type (for example, OC-48
single-mode fiber) and sent out through the transponder.
The long distances made possible by advances in technologies such as optical amplifiers, dispersion
compensators, and new fiber types, resulted in the initial deployment of DWDM technology in the
long-haul transoceanic and terrestrial networks. Once these technologies became commercially viable
in the long-haul market, it was the next logical step to deploy them in the metropolitan area and,
eventually, in the access networks using hybrid architectures of fiber and coaxial media.
We begin the following discussion by considering the main transport and network technologies used in
metropolitan networks. We then briefly explore some of the potential applications for DWDM in these
networks and examine the topologies that can be deployed. We continue by examining the protection
mechanisms and schemes that are available to ensure reliability, and we conclude with a look at the
future of optical networking.
SONET/SDH
SONET/SDH has been the foundation for MANs over the last decade, serving as the fundamental
transport layer for both TDM-based circuit switched network and most overlay data networks. While
SONET/SDH has evolved into a very resilient technology, it remains fairly expensive to implement.
Inherent inefficiencies in adapting data services to the voice-optimized hierarchy and an inflexible
multiplexing hierarchy remain problematic. More importantly, capacity scaling limitations—OC-768
may be the practical limit of SONET/SDH—and unresponsiveness to bursty IP traffic make any
TDM-based technology a poor choice for the future.
ATM
Many service providers favor ATM because it can encapsulate different protocols and traffic types into
a common format for transmission over a SONET infrastructure. Meanwhile the data networking world,
which is overwhelmingly IP-oriented, favors packet over SONET (POS), which obviates the costly ATM
intermediate layer. Advancements in IP, combined with the scaling capacity of gigabit and multigigabit
routers, make it possible to envisage an IP-based network that is well suited for carrying primarily data
traffic, and secondarily voice.
Nevertheless ATM remains strong in the metropolitan area. It can accommodate higher speed line
interfaces and provide managed virtual circuit services while offering traffic management capabilities.
Thus ATM edge devices are commonly used to terminate traffic, including VoIP, DSL, and Frame Relay.
Gigabit Ethernet
Gigabit Ethernet (GE) is a proven technology for easy migration from and integration into traditional
Ethernet. It is relatively inexpensive compared to other technologies that offer the same transmission
rate, but does not provide quality of service (QoS) or fault tolerance on its own. When confined to
point-to-point topologies, collisions and carrier sense multiple access (CSMA) are not of concern,
resulting in more effective use of the full bandwidth. Because the optical physical layer can support
much longer distances than traditional Category 5 cable, Gigabit Ethernet over fiber (1000BASE-LX,
for example) can be extended into the wide-area realm using DWDM.
The latest advancement in Ethernet technology, 10 Gigabit Ethernet, is being driven by a need to
interconnect Ethernet LANs operating at 10, 100, or 1000 Mbps. Ten Gigabit Ethernet can be used for
aggregating slower access links, in the backbone networks, and for WAN access. Using 1550-nm serial
lasers, distances of 40 to 80 km (25 to 50 mi) are possible with 10 Gigabit Ethernet over standard SM
fiber. With such technology, service providers can build simple Ethernet networks over dark fiber
without SONET or ATM and provision high-speed 10/100/1000 Mbps services at very low cost. In
addition, a very short reach (VSR) OC-192 interface can be used to connect 10 Gigabit Ethernet to
DWDM equipment over MM fiber.
Ethernet offers the technical advantages of a proven, adaptable, reliable, and uncomplicated technology.
Implementations are standard and interoperable, and cost is much less than SONET or ATM.
Architecturally, Ethernet’s advantage is its emerging potential to serve as a scalable, end-to-end solution.
Network management can also be improved by using Ethernet across the MAN and WAN.
IP
Clearly, as traditional circuit-switched services migrate to IP networks and data grows, networks must
evolve to accommodate the traffic. However, IP may need to become as complex as ATM to replace its
functionality. Thus, both ATM and IP are candidates for transport directly over DWDM. In either case,
the result is simplified network infrastructure, lower cost due to fewer network elements and less fiber,
open interfaces, increased flexibility, and stability. The question is, in which format will IP travel over
an optical network: IP over ATM over SONET, IP over SONET (as POS), or IP over Gigabit Ethernet or
10 Gigabit Ethernet? (See Figure 3-1.)
Figure 3-1 Data Link and Network Protocols over the Optical Layer
IP
GE/10GE
FC
ATM
SONET
49368
Optical physical layer
Fibre Channel
Fibre Channel is the predominant data link technology used in storage area networks (SANs). See the
“Storage Area Networks” section on page 3-5. Fibre Channel is an economical replacement for the Small
Computer System Interface (SCSI) protocol as a high-speed interface for applications such as data
backup, recovery, and mirroring. Fibre Channel interfaces are available at 100 MBps today; 200 MBps
interfaces should be available in the near future, and 400 MBps interfaces are in testing.
Note By convention, transfer rates for storage are specified in megabytes per second (MBps).
Fibre Channel comes without the very short distance limitations of SCSI; it also avoids the termination
restrictions of SCSI because each node acts as an optical repeater. Fibre Channel can be implemented in
a point-to-point, arbitrated loop, or mesh topology using a switch. As shown in Figure 3-1, Fibre
Channel, like other protocols, can be carried directly over the optical layer using DWDM.
Outer ring
Inner ring
Outer ring
control
data
Outer ring
control
Inner ring
data
Inner ring
49369
FDDI
FDDI is at this point a legacy technology. Having served a need at one time, it has been replaced by more
advanced technologies. Although FDDI is capable of scaling to the metropolitan area, it is also a shared
media technology with a relatively low capacity by current standards. This limitation, along with falling
availability of FDDI interfaces on network equipment, is causing FDDI to be replaced by Gigabit
Ethernet, or ATM. Nevertheless, it is also a protocol that can be transparently transported over the optical
layer using DWDM.
Local area
network
Server Server Server Server
Storage area
network
Tape Tape
Disk array Disk array
49370
A number of types of interfaces have been used to connect servers to devices in a SAN. The most
prevalent is IBM’s Enterprise System Connection (ESCON), a 17-MBps half-duplex protocol over fiber.
Fibre Channel, on which IBM’s FICON is based, is also frequently employed in SANs and has a much
higher capacity than ESCON (see the “Fibre Channel” section on page 3-3.) Both technologies, however,
have significant distance limitations. For example, the standard maximum distance without repeaters is
around 3 km (1.9 mi) for half duplex ESCON and around 10 km (6.2 mi) for full duplex 100-MBps Fibre
Channel. There is performance degradation as distances increase beyond these numbers.
This distance limitation can be overcome by transporting data between one or more enterprise locations
and one or more SANs over the optical layer using DWDM. In Figure 3-4, for example, the distance
separating the enterprise location and the SAN sites can be greatly extended. Access to the ring is by
way of “satellite” OADMs with Fibre Channel or ESCON interfaces at each SAN location (one of these
could provide data mirroring). These interfaces can also support Sysplex Timer and Sysplex Coupling
Link interfaces, used in IBM environments for distributing loads across the members of a server
complex.
Enterprise
Primary Standby
SAN SAN
49371
In addition to overcoming distance limitations, DWDM can also reduce fiber requirements in SANs.
Both ESCON and FICON require a pair of fibers for every channel. By multiplexing these channels over
DWDM transport, significant savings can be realized.
SONET
ADM
SONET SONET
ADM ADM
ADM
SONET
49372
The third option is shown in Figure 3-6. By using DWDM to increase the capacity of the existing ring,
one fiber can essentially act as many.
SONET
ADM
SONET
ADM
SONET SONET
ADM ADM
ADM ADM
49373
SONET SONET
In a second type of scenario, DWDM can be used to remove an entire class of equipment, the SONET
ADMs. This change, which might constitute a second phase of SONET migration, allows routers and
other devices to bypass SONET equipment and interface directly to DWDM, while simplifying traffic
from IP/ATM/SONET to POS to eventually IP directly over the optical layer (see Figure 3-7).
OC-48
OC-12 OC-12
OC-48 OC-48
49374
In this phase of migration, end user sites are served by OADMs rather than SONET ADMs. In this way
DWDM rings and mesh networks can eliminate the increased cost and complexity of introducing more
SONET elements into the network to meet demand. The advantage here for carriers is the ability to offer
bit-rate-independent services, freeing them from the DS1/DS3/OC-n framework. Such a scheme would
also allow enterprise LAN access to be extended across the MAN or WAN without a SONET
infrastructure.
A further advantage in migrating from SONET to the optical layer is that protection and restoration
becomes less susceptible to failure of electronic components; a common survivability platform for all
network services is created, including those without built-in protection.
Point-to-Point Topologies
Point-to-point topologies can be implemented with or without OADM. These networks are characterized
by ultra-high channel speeds (10 to 40 Gbps), high signal integrity and reliability, and fast path
restoration. In long-haul networks, the distance between transmitter and receiver can be several hundred
kilometers, and the number of amplifiers required between endpoints is typically less than 10. In the
MAN, amplifiers are often not needed.
Protection in point-to-point topologies can be provided in a couple of ways. In first generation
equipment, redundancy is at the system level. Parallel links connect redundant systems at either end.
Switchover in case of failure is the responsibility of the client equipment (a switch or router, for
example), while the DWDM systems themselves just provide capacity.
In second generation equipment, redundancy is at the card level. Parallel links connect single systems at
either end that contain redundant transponders, multiplexers, and CPUs. Here protection has migrated to
the DWDM equipment, with switching decisions under local control. One type of implementation, for
example, uses a 1 + 1 protection scheme based on SONET Automatic Protection Switching (APS). See
Figure 3-8.
APS APS
OADM
APS APS
49375
Ring Topologies
Rings are the most common architecture found in metropolitan areas and span a few tens of kilometers.
The fiber ring might contain as few as four wavelength channels, and typically fewer nodes than
channels. Bit rate is in the range of 622 Mbps to 10 Gbps per channel.
Ring configurations can be deployed with one or more DWDM systems, supporting any-to-any traffic,
or they can have a hub station and one or more OADM nodes, or satellites (see Figure 3-9). At the hub
node traffic originates, is terminated and managed, and connectivity with other networks is established.
At the OADM nodes, selected wavelengths are dropped and added, while the others pass through
transparently (express channels). In this way, ring architectures allow nodes on the ring to provide access
to network elements such as routers, switches, or servers by adding or dropping wavelength channels in
the optical domain. With increase in number of OADMs, however, the signal is subject to loss and
amplification can be required.
Hub
OADM OADM
λ1, λ2,....λn
λ9-12
λ1-4
λ1, λ2,....λn OADM
λ1, λ2,....λn
49376
λ5-8
Candidate networks for DWDM application in the metropolitan area are often already based on SONET
ring structures with 1 + 1 fiber protection. Thus schemes such as Unidirectional Path Switched Ring
(UPSR) or Bidirectional Line Switched Ring (BLSR) can be reused for DWDM implementations.
Figure 3-10 shows a UPSR scheme with two fibers. Here, hub and nodes send on two counter-rotating
rings, but the same fiber is normally being used by all equipment to receive the signal; hence the name
unidirectional. If the working ring should fail, the receiving equipment switches to the other pair.
Although this provides full redundancy to the path, no bandwidth reuse is possible, as the redundant fiber
must always be ready to carry the working traffic. This scheme is most commonly used in access
networks.
Hub
UPSR
OADM
OADM
49377
Other schemes, such as Bidirectional Line Switched Ring (BLSR), allow traffic to travel from the
sending to the receiving node by the most direct route. Because of this, BLSR is considered preferable
for core SONET networks, especially when implemented with four fibers, which offers complete
redundancy.
Mesh Topologies
Mesh architectures are the future of optical networks. As networks evolve, rings and point-to-point
architectures will still have a place, but mesh promises to be the most robust topology. This development
will be enabled by the introduction of configurable optical cross-connects and switches that will in some
cases replace and in other cases supplement fixed DWDM devices.
From a design standpoint, there is a graceful evolutionary path available from point-to-point to mesh
topologies. By beginning with point-to-point links, equipped with OADM nodes at the outset for
flexibility, and subsequently interconnecting them, the network can evolve into a mesh without a
complete redesign. Additionally, mesh and ring topologies can be joined by point-to-point links (see
Figure 3-11).
Ring
Mesh Point-to-point
49378
DWDM mesh networks, consisting of interconnected all-optical nodes, will require the next generation
of protection. Where previous protection schemes relied upon redundancy at the system, card, or fiber
level, redundancy will now migrate to the wavelength level. This means, among other things, that a data
channel might change wavelengths as it makes its way through the network, due either to routing or to
a switch in wavelength because of a fault. The situation is analogous to that of a virtual circuit through
an ATM cloud, which can experience changes in its virtual path identifier (VPI)/virtual channel identifier
(VCI) values at switching points. In optical networks, this concept is sometimes called a light path.
Mesh networks will therefore require a high degree of intelligence to perform the functions of protection
and bandwidth management, including fiber and wavelength switching. The benefits in flexibility and
efficiency, however, are potentially great. Fiber usage, which can be low in ring solutions because of the
requirement for protection fibers on each ring, can be improved in a mesh design. Protection and
restoration can be based on shared paths, thereby requiring fewer fiber pairs for the same amount of
traffic and not wasting unused wavelengths.
Finally, mesh networks will be highly dependent upon software for management. A protocol based on
Multiprotocol Label Switching (MPLS) is under development to support routed paths through an
all-optical network. In addition, network management will require an as-yet unstandardized channel to
carry messages among the network elements.
Interoperability Issues
Because DWDM uses specific wavelengths for transmission, the wavelengths used must be the same on
either end of any given connection. Toward this end, the ITU has standardized on a grid with spacings
of 100 GHz (see Table 2-1 on page 2-14). However, vendors may use wider spacing, sometimes at
200 GHz, or narrower. In addition, different vendors who do use the same grid may not use the same
lambda numbering scheme. That is, lambda 1 on vendor A’s equipment may be assigned a different
wavelength from lambda 1 on vendor B’s equipment. Hence, it is important to be aware of the potential
interoperability problems posed by different grid alignments.
Other interoperability issues include power levels, inter- and intra-channel isolation, PMD tolerances,
and fiber types. All these contribute to the challenges of transmission between different systems at
Layer 1.
Future of DWDM
DWDM will continue to provide the bandwidth for large amounts of data. In fact, the capacity of systems
will grow as technologies advance that allow closer spacing, and therefore higher numbers, of
wavelengths. But DWDM is also moving beyond transport to become the basis of all-optical networking
with wavelength provisioning and mesh-based protection. Switching at the photonic layer will enable
this evolution, as will the routing protocols that allow light paths to traverse the network in much the
same way as virtual circuits do today.
These and other advances are converging such that an all-optical infrastructure can be envisioned.
Figure 3-12 shows an example of such an infrastructure, using mesh, ring, and point-to-point topologies
at the optical layer to support the needs of enterprise, metropolitan access, and metropolitan core
networks.
POP To/from
Enterprise national
Data center backbone
Residential access
DWDM rings
DWDM mesh
DWDM bus
49379
competition 1-4
A
absorption 2-8
access networks
D
about 1-2 demultiplexers
demand 1-11 See multiplexers/demultiplexers
additional reading vii dense wavelength division multiplexing
arrayed waveguide gratings 2-18 See DWDM
Asynchronous Transfer Mode dispersion
See ATM causes 2-10
ATM chromatic 2-10
statistical multiplexing 1-6 definition 2-8
technology in MANs 3-1 modal 2-6
attenuation polarization mode 2-11
causes 2-8 dispersion-shifted fiber
definition 2-8 See DSF
DPT 3-3
DSF 2-7, 2-12
B
DWDM
bandwidth advantages 1-12, 1-16
demand 1-3 considerations for deployment 3-13
drivers 1-3 future of 3-14
options for increasing 1-5 requirements in MAN 1-12
bit rate, increasing topologies 3-9 to 3-12
DWDM 1-14 value in MAN 1-11
TDM 1-5 See also WDM
DWDM components
demultiplexers 2-16
C
light detectors 2-13, 2-15
chromatic dispersion light sources 2-13
effects 2-10 multiplexers 2-16
material 2-10 optical amplifiers 2-15
waveguide 2-10 optical fibers 2-5
overview 2-5
G
passive 1-15
transponders 2-21 Gigabit Ethernet
DWDM systems DWDM interfaces 2-20
channel density 2-3 technology in MANs 3-2
components 2-5, 2-13 to 2-21
development 2-2
I
enabling technologies 2-4
functions 2-3 Internet Protocol
interfaces 2-20 See IP
interoperability 3-14 interoperability 3-14
operation 2-21 IP
Dynamic Packet Transport technology in MANs 3-2
See DPT traffic growth 1-3
ITU laser grid 2-14
E
L
EDFAs
operation 2-16 lasers
performance parameters 2-16 designs 2-14
erbium-doped fiber amplifiers ITU grid 2-14
See EDFAs modulating 2-14
ESCON operation 2-13
DWDM interfaces 2-20 LEDs 2-13
in SANs 3-6 light emitting diodes
See LEDs
link loss budgets
F
See optical power budgets
FDDI 3-4 long-haul networks
Fiber Distributed Data Interface about 1-2
See FDDI MANs, compared 1-2
fiber optic transmission 2-1
Fibre Channel
M
DWDM interfaces 2-20
in SANs 3-6 MANs
technology in MANs 3-3 about 1-2
four-wave mixing 2-11 access segment 1-3
applications in 3-5